Cookie Run: Braverse Card List Filters Are Missing Cards

Open the official CookieRun: Braverse card list, scroll to the Effect block, and tick the chip marked On play. You get an empty page.

There are 268 entries in that same database whose skill text prints 【On Play】. The chip is not finding a small number of them, or the wrong ones. It finds zero.

That is not a one-off. The card list matches most of its filter chips as exact, case-sensitive string comparisons, and the underlying data spells several of those values more than one way. Every number below was checked against the publisher's own card feed at cookierunbraverse.com/data/json/cardList_en.json — 2,274 entries as of September 12, 2026 — and then reproduced against the live site by URL.

The two-minute reproduction

The filter chips write themselves into the query string, so you can check any of this without clicking. These five URLs are the ones I ran:

The Dragon row is the control, and it is clean: the trait lives in one field, the chip's label matches what is stored, and the count is right down to the card. The comparison itself works fine.

The Equip row looks like a second control and is not, which makes it the most instructive row in the table. The chip's spelling is correct — 【Equip】 is exactly how the cards print it — and it still returns 13 of the 15 entries that print the tag, because the Effect block reads a card's skill text field and nothing else. Two cards print 【Equip】 in their attack text instead, and the chip never looks there. A correctly-spelled chip is still not a complete list.

The last three rows are the louder failure: the same code hitting values that are spelled differently in the data than on the chip.

What each filter actually matches

The comparison logic ships in the page's own JavaScript bundle (/_nuxt/CgiSeUy6.js), one branch per filter. Written out in English:

Two things to read off that table. First, COST is the only filter that is case-insensitive — it lower-cases both sides and then asks for a substring rather than an exact match, which is why ticking Red returns the 319 plain RED cards, the 98 RED MIX cards and the eight multi-color entries that list RED among their costs. Every other text filter compares raw. Second, the right-hand column is not a count of bugs; it is a count of entries that no chip in that block can return, for any reason. The sections below split those out.

The Effect filter: two chips return nothing

The Effect block reads the bracketed ability tags out of a card's skill text and asks whether the chip's label is one of them, character for character. Braverse prints those tags in title case. Two of the six chips are not written in title case.

Read the last column as the tax every Effect chip pays regardless of spelling: those entries print the tag on the card, but print it in the attack text, so no chip returns them. Activate is the worst of it — the chip finds 355 and 29 more exist.

For the two broken chips, a lower-case p and a lower-case t are the whole story. On Play is the second most common ability tag in the game — more common than Blocker, Equip and Your Turn combined — and it is the one the chip cannot see.

The Effect block is also missing chips for tags that do exist. Checked across all 2,274 entries, the thirteen bracketed tags that appear in skill text are Activate, On Play, Once Per Turn, Arena, EXTRA, Blocker, Your Turn, Ancient, Special Play, Awaken, Equip, Skill and Beast. Six of those have a chip. There is no chip for Special Play (17 entries) or Awaken (14 entries) — or for Arena, EXTRA, Ancient, Skill and Beast as skill-text tags — so a list of any of them has to be assembled from the card text rather than the filter. Our Special Play guide covers what that mechanic does.

Put together, 462 entries print at least one bracketed tag somewhere on the card and are returned by none of the six Effect chips. Three separate things produce that number: 283 are On Play and Your Turn casualties whose spelling the chip cannot match; 37 print a tag the block does have a chip for, but print it only in their attack text; and 142 print only tags that have no chip at all, 96 of them 【Arena】.

The Keyword filter: ANCIENT is short 18 cards

Braverse's four tribal traits live in a separate field, and that field has been populated inconsistently across sets.

The 18 entries filed as title-case Ancient are not obscure. They are the five founding Ancient Cookies from Booster 3 [Age of Heroes and Kingdoms] — Hollyberry, Golden Cheese, White Lily, Pure Vanilla and Dark Cacao — at both their Ultra Rare and Secret Ultra Rare printings, plus the Golden Cheese and Dark Cacao cards from Booster 8.

You can watch it happen in one comparison. ?keyword=ANCIENT&rare=UR returns 4 cards. ?search=ancient&rare=UR returns 12. The four the chip finds are real. Seven of the eight it drops carry the Ancient trait — all five Booster 3 Ancients plus the Booster 8 Golden Cheese and Dark Cacao — and the eighth, the Stage card Age of Heroes and Kingdoms at BS3-121, simply has the word in its text.

The Arena side is smaller but stranger: seven promo printings are filed as ARENA in caps, and one is filed as Arema. That misspelling is the Cream Puff Cookie promo at BS7_093@2 — the same entry our Festival Arena set list flagged when it was counting the Arena pool, and one of five entries in the database whose card number uses an underscore instead of a hyphen.

If you want the real trait totals rather than the filtered ones: 340 Arena entries, 58 Ancient, 42 Beast, 19 Dragon. Our keywords and abilities explainer covers what each trait does; the counts above are what the database holds today.

HP, and the three different plus signs

The HP chips run 1 to 6 and compare numbers. Fifteen entries store an HP value that is not a number as far as that comparison is concerned, because Awakened forms print an HP bonus rather than an absolute — and the database has typed that bonus three different ways.

Golden Cheese Cookie BS8-027 is the tidiest demonstration of this, because the split runs through a single card number. It has six printings. Three of them store +2 and three store ＋2. Search the site for BS8-027 and you get all six; add the HP 2 chip and you get three. Same card, same printed HP, half the results.

Three more printings — all of Pure Vanilla Cookie at BS9-088 — store an HP of 0, and there is no HP 0 chip, so they are unreachable from that block too.

Damage stops at 4, Level skips a promo

The Damage chips offer 1, 2, 3 and 4. Across the database the first damage value on a card runs 1 (406 entries), 2 (492), 3 (451), 4 (79), 5 (4) and 7 (2). The six entries above 4 are the two printings each of Elder Faerie Cookie BS3-060 and Avatar of Destiny BS11-091 at 5 damage, and the two printings of Dark Enchantress Cookie BS11-116 at 7. The game's hardest-hitting card is the one the damage filter cannot list — and because its HP is that subscript plus sign, it is out of reach from the HP block as well.

The Level chips offer 1, 2, 3 and 5, which is correct: there is no Level 4 card in the database, and the only two Level 5 entries are the base and @1 printings of that same EXTRA Dark Enchantress Cookie. One entry misses: Spinach Cookie P-131 stores its level as a fullwidth １, so it answers to no Level chip and, thanks to its fullwidth ２ HP, no HP chip either. It is still findable under Activate and Once Per Turn.

Card Color: 68 entries have no color

Card Color compares as a case-sensitive substring, and where a color is stored it returns the card. The problem is the 68 entries where there is nothing to compare against.

Sixty-two of them store an empty color, and all 62 are the plain-MIX Cookies — the ones whose cost is any color rather than a specific one. Thirty-four are from Booster 1 [BRAVE BEGINNING] alone, including Melon Bun Cookie, Walnut Cookie, Apple Cookie and Pizza Cookie. That may well be deliberate: a colorless card arguably has no color to file. The practical effect is still that there is no chip — not in Card Color and not in COST — that returns them, so they drop out of any color-filtered list you build.

The other six are harder to read as intentional. They store the literal text "null" as their color:

Three of the six have a base printing in the same database that carries a real color. Frost Queen Cookie is Blue at BS8-083 and colorless-by-typo at BS8-083@2, which is its Secret Ultra Rare.

COST has a related gap of its own: three promo printings of Wind Archer Cookie (BS2_058@2, @3 and @4) store their energy as PULPLE, so the Purple cost chip does not return them.

Rarity, Type, and the toggles

Two smaller ones, both exact-string comparisons:

• Rarity offers eleven codes, C through P. One entry does not use a bare code: Blue Slushy Cookie ST8-007, from the Guardian Winds starter deck, stores its rarity as C (홀로) — "C (holo)" with the Korean word left in. It answers to no rarity chip.
• Card Type offers six options: COOKIE, ITEM, TRAP, STAGE, FLIP and EXTRA. The database also contains three cards typed NPC — TBD Delivery Cookie, TBD Office Clerk Cookie and TBD Mender Cookie, all BS6-028 to BS6-030 from Operation Timeguard. There is no NPC chip, so there is no way to list them from the Card Type block.

The toggles below the chips work differently and are worth understanding, because they are on by default and only ever remove cards. Untick extra and EXTRA-deck cards drop out; untick Duplicated Card and each @ printing collapses to the lowest-numbered version of its card number, which is genuinely useful when you are counting distinct cards rather than printings. Untick Banned Card and nothing changes at all — the ban flag is 0 on all 2,274 entries, which is the same thing our banned and restricted list found when it went looking: the English B&R list is published as a policy notice and is not reflected in the card data.

One display note that is not a filter problem but will confuse you on a card's detail page: the rarity word is a separate field from the rarity code, and it has its own typos. Seven entries in Booster 9 display ULTLA RARE and one Booster 3 Uncommon displays UOMMON — both first flagged in our card numbers explainer — and thirteen more carry a stray line break in front of the rarity word: ten before SECRET RARE, two before SECRET ULTRA RARE and one before EXTRA RARE. The code underneath is correct in each case, which is why Rarity filtering still works on them.

How to search around it

The free-text box at the top of the card list does not share the chips' problem. Its matcher lower-cases both sides, collapses runs of whitespace, and walks every field on the card record looking for a substring — card number, name, type, rarity, keyword, skill text, attack text and product name all at once. That is why ?search=ancient finds all 58 Ancient-trait entries while the ANCIENT chip finds 40.

So, in order of how much work they are:

1. For a trait or an ability, type it into the search box instead of using the chip. Case does not matter. "on play", "your turn" and "ancient" all work.
2. Combine the search box with the chips that do work. Rarity, Type and Damage are exact and reliable within their listed options, so ?search=your+turn&rare=UR is a sound query even though ?skill=Your+turn&rare=UR returns nothing.
3. Check both spellings when you are counting. For Ancient, search rather than filter. For Arena, the chip misses 8 of 340. For Beast and Dragon the chip and the data agree, so the chip is fine.
4. If you need the whole pool, take the feed. cookierunbraverse.com/data/json/cardList_en.json is a single public JSON file holding the English entry list — 2,274 rows on September 12, 2026 — with the fields behind each chip on the record. It is what the page itself loads, and it is where an exact count comes from.

A caveat on the search box: because it matches across all fields, broad words over-match. Searching "ancient" returns 95 entries, not 58 — the extra 37 are entries with "Ancient" somewhere in their name or text that do not carry the trait. That is the correct trade. The chip is precise and incomplete; the search box is complete and imprecise. For counting, filter the JSON.

What this does not mean

Some boundaries, because a data problem is easy to inflate into a game problem:

• No card is affected. This is a website search index, not a rules document. Pure Vanilla Cookie has the Ancient trait whether or not a chip returns it, and the printed card is what a judge reads.
• Nothing here is a legality claim. Deck legality comes from Devsisters' policy notices, not from the card list's filters or its ban flag.
• I checked the English site only. The Korean and Asia card lists run the same page code but are served different data files, and I have not verified whether the same spellings appear there.
• This is a snapshot. All counts were pulled on September 12, 2026, from a database that gained 31 promo entries between September 4 and September 7. Devsisters can fix any of this in a single data edit, and the chip counts will move when they do.

If you are building a list of anything — a trait, an ability, a color — start from the feed or the search box, and treat the chips as a fast filter rather than a complete one.